generated: '2026-08-12' method: generated source: openapi/ + conventions/ + errors/ + data-model/ note: >- Agent Skills authored by API Evangelist against the derived OpenAPI documents in openapi/. Every operationId referenced below exists in a spec in this repository, and every spec was derived from the live https://www.exentis-group.com/wp-json/ discovery document. Exentis Group publishes no skills, no AGENTS.md and no developer documentation of any kind — nothing here is provider-authored. skills: - name: exentis-group-track-press-releases file: exentis-group-track-press-releases.md api: exentis-group:exentis-group-posts-api operations: - listPosts - getPost - listCategories summary: >- Watch the company's announcements over time — latest items by language, correct header-based pagination through the 371-item archive, and change detection via modified_after. - name: exentis-group-research-technology-pages file: exentis-group-research-technology-pages.md api: exentis-group:exentis-group-search-api operations: - searchSite - getPage - listPages - getBlogPost - listLanguages summary: >- Answer technology, materials, applications, certifications and sustainability questions by resolving a query to real ids through search, then fetching structured page and blog bodies. Carries the subtype-to-collection routing rule that is the easiest thing to get wrong here. - name: exentis-group-harvest-media-assets file: exentis-group-harvest-media-assets.md api: exentis-group:exentis-group-media-api operations: - listMedia - getMediaItem - getPage summary: >- Locate brand, product and trade-fair imagery in the 1,155-item media library, filter by MIME type, and select the right generated rendition instead of pulling originals. exclusions: - surface: /wp/v2/users entity: Author reason: personal-data detail: >- DELIBERATE OMISSION, not an oversight. The users collection returns four real person records anonymously — name, slug, description, avatar URLs and ACF fields — and Post, Page and MediaItem all carry an `author` id that resolves into it. No skill wraps this endpoint, no MCP tool is proposed for it, and no individual is named anywhere in this repository. It is documented structurally in data-model/exentis-group-data-model.yml so the exposure is on record without shipping a working harvester. Every skill above also forbids `_embed`, which would inline the same records as a side effect of an otherwise innocent call. - surface: /wp-json/mcp/mcp-adapter-default-server reason: auth-gated detail: >- The MCP server returns 401 to anonymous tools/list and initialize, so no skill can name a tool it exposes. See mcp/exentis-group-mcp.yml. - surface: write operations reason: no-public-credentials detail: >- POST/PUT/DELETE on any route requires a WordPress application password issued only to Exentis Group site staff. All skills are read-only by construction. operating_constraints: robots_txt: >- https://www.exentis-group.com/robots.txt bans roughly 35 named AI and agent crawlers by user-agent (Claude-User, CCBot, OAI-SearchBot, meta-externalagent, Bytespider, MistralAI-User, Diffbot, cohere-ai, AI2Bot and others) and applies `Disallow: /*?` to everyone else. Every filtered REST call carries a query string. Any scheduled use of these skills needs an explicit exemption or the site owner's agreement. rate_limits: >- None published and no rate-limit headers returned, so there is no backoff signal. Skills instruct self-throttling. licensing: >- Public readability is not a content licence. See https://www.exentis-group.com/en/terms-of-use/.